Cladogenesis and Reticulation in Cuscuta Sect. Denticulatae (Convolvulaceae)

Home , Cladogenesis, Cuscuta compacta, Cuscuta denticulata, Cuscuta pacifica

Organisms Diversity & Evolution (2018) 18:383–398 https://doi.org/10.1007/s13127-018-0383-5

ORIGINAL ARTICLE

Cladogenesis and reticulation in Cuscuta sect. Denticulatae (Convolvulaceae)

Miguel A. García1,2 & SašaStefanović1 & Catherine Weiner3 & Magdalena Olszewski3 & Mihai Costea3

Received: 5 May 2018 /Accepted: 11 October 2018 /Published online: 28 October 2018 # The Author(s) 2018

Abstract As traditionally circumscribed, Cuscuta sect. Denticulatae is a group of three parasitic plant species native to the deserts of Western USA (Cuscuta denticulata, Cuscuta nevadensis) and the central region of Baja California, Mexico (Cuscuta veatchii). Molecular phylogenetic studies confirmed the monophyly of this group and suggested that the disjunct C. veatchii is a hybrid between the other two species. However, the limited sampling left the possibility of alternative biological and methodological explanations. We expanded our sampling to multiple individuals of all the species collected from across their entire geographical ranges. Sequence data from the nuclear and plastid regions were used to reconstruct the phylogeny and find out if the topological conflict was maintained. We obtained karyotype information from multiple individuals, investigated the morphological variation of the group thorough morphometric analyses, and compiled data on ecology, host range, and geographical distribution. Our results confirmed that C. veatchii is an allotetraploid. Furthermore, we found previously unknown autotetraploid population of C. denticulata, and we describe a new hybrid species, Cuscuta psorothamnensis. We suggest that this newly discovered natural hybrid is resulting from an independent (and probably more recent) hybridization event between the same diploid parental species as those of C. veatchii. All the polyploids showed host shift associated with hybridization and/or polyploidy and are found growing on hosts that are rarely or never frequented by their diploid progenitors. The great potential of this group as a model to study host shift in parasitic plants associated with recurrent allopolyploidy is discussed.

Keywords Host shift . Host range . Hybridization . Polyploidy . Parasitic plant . Speciation

Introduction cosmopolitan but the majority of species are native to North and South America, belonging to Cuscuta subg. Grammica,the Cuscuta (dodders) is a plant genus of nearly 200 species of largest infrageneric group that includes more than 150 species stem parasites (Yuncker 1932;Garcíaetal.2014; Costea et al. (Costea et al. 2015a). Dodders occur in a great variety of hab- 2015a) that has evolved within Convolvulaceae (reviewed by itats, from temperate to tropical, desert to riparian, littoral to Stefanović and Olmstead 2004, 2005). The genus is nearly high mountains, grasslands, forests, saline, and disturbed hab- itats. Similarly to other parasitic plants, dodders act as keystone species in their ecosystems (Press and Phoenix 2005;Graffis Miguel A. García and Saša Stefanović contributed equally to this work. and Kneitel 2015). Approximately 15–20 Cuscuta spp. world- Electronic supplementary material The online version of this article wide are agricultural and horticultural pests (Dawson et al. (https://doi.org/10.1007/s13127-018-0383-5) contains supplementary 1994; Costea and Tardif 2006), and in most countries, control material, which is available to authorized users. and quarantine measures target the genus as a whole, ignoring * Miguel A. García the fact that some species may be endangered or even threat- [email protected] ened with extinction (Costea and Stefanović 2009). Stefanović et al. (2007) noted several cases of conflict between plastid- and nuclear-derived phylogenies indicative of 1 Department of Biology, University of Toronto Mississauga, Mississauga, ON L5L 1C6, Canada possible reticulation in Cuscuta subg. Grammica.Tofurther investigate the origin of these conflicts, Stefanović and Costea 2 Royal Botanic Gardens Kew, Richmond, Surrey TW9 3AE, UK 3 (2008)expandedtheirtrnL-F and nrITS matrices through ad- Department of Biology, Wilfrid Laurier University, dition of multiple sequences from 105 species from this Waterloo, ON N2L3C5, Canada 384 García M.A. et al. subgenus and found five cases of species with a probable hybrid origin. A series of statistical tests by Stefanović and Costea (2008) showed that alternative hypotheses to explain discordant gene topologies, such as incomplete lineage sorting, undetected paralogy, or horizontal gene transfer, were not supported. All these cases in subg. Grammica were confirmed and three additional ones were detected using rbcL and nrLSU sequences in a broader phylogenetic context with rep- resentatives of the entire genus (García et al. 2014). Several cases of topological incongruence were also found for Cuscuta subg. Cuscuta by García and Martín (2007), but the origin of the conflicts was not further investigated. Costea and Stefanović (2010) discovered that at least four independent hybridization events had occurred in the evolution of Cuscuta sect. Umbellatae (subg. Grammica), two more for this section than previously detected. More recently, a detailed study by Costea et al. (2015b)onCuscuta sect. Cleistogrammica (subg. Grammica) showed that the worldwide invasive weed Cuscuta campestris Yunck (1932) has two divergent groups of nrITS ribotypes. Both of these disparate ribotypes are topologi- cally incongruent with the plastid trnL-F phylogeny, in aggregate suggesting the hybrid origin of C. campestris (Costea et al. 2015b). Another striking case of topological discordance was found within Cuscuta sect. Denticulatae (Stefanović and Costea 2008;Garcíaetal.2014). This group of species in subg. Grammica is well characterized morphologically by the radicular end of the embryo spherically enlarged in a ball-like structure that increases in volume during seed maturation (Costea et al. 2015a). Such a feature is not present in any other clade of Cuscuta and this synapomorphy is thought to be an adaptation for seed germination on the host while the fruit is still enclosed by the perianth (vivipary). The enlarged embryo probably stores nutrients and water as an adaptation to germination in desert environments (Costea et al. 2005). Section Denticulatae includes three species distributed in Fig. 1 Distribution of Cuscuta sect. Denticulatae species across their Western USA (Cuscuta denticulata Engelm.; Cuscuta geographic ranges in western North America. Potential extent of nevadensis I.M.Johnst.) and the Central Desert of Baja distribution for C. denticulata is outlined and that of C. nevadensis is shaded. Approximate positions of sampling sites used in this study are California in Mexico (Cuscuta veatchii Brandegee) (Fig. 1). indicated (for details, see Appendix 1). Circles (solid and open) represent Of the three species, C. denticulata has the broadest geograph- sampling sites for populations of C. denticulata, squares (solid and open) ical distribution and host preference, whereas C. nevadensis represent those of C. nevadensis, triangles those of C. veatchii, while X has narrower geographical and host ranges, occurring sympat- symbols stand for the newly described species, C. psorothamnensis. Encircled symbols represent material obtained from herbaria; all others rically with C. denticulata (Costea et al. 2005). Cuscuta are sampled directly in the field, including multiple individuals per pop- veatchii has a disjunct distribution and grows only on ulation. Solid and open symbols correspond to different haplo- and Pachycormus discolor (Benth.) Coville (Anacardiaceae). ribotypes of C. denticulata and C. nevadensis (see text for details) The phylogenetic analyses by Stefanović and Costea (2008) placed C. veatchii in a clade with C. denticulata on the nrITS tree, whereas this species was resolved with C. nevadensis on to find statistical difference between optimal and constrained the plastid trnL-F tree, in both cases with 100% bootstrap trees among plastid and nuclear data. Also, in that study, the support. The authors concluded that this topological incongru- sampling was limited to two to three individuals per species, ence, together with the life history, ecological, and biogeo- and because the clade included only three taxa and a root, a graphical data, was consistent with a hybrid origin of topological distortion such a nearest-neighbor interchange C. veatchii. However, some alternative topology tests failed (NNI) could not be ruled out, leaving ancestral polymorphism Cladogenesis and reticulation in Cuscuta sect. Denticulatae (Convolvulaceae) 385 or incomplete lineage sorting as viable alternative explana- population sequence variation of DNA markers, two to five tions for the discordance between nuclear and plastid trees. individuals were sampled per locality in our field trips. Within In the present study, we further investigate the hybridiza- a locality, multiple specimens were collected growing on sep- tion hypothesis for the origin of C. veatchii by expanding our arate host plants (> 1 m apart from each other). Despite the sampling to multiple individuals for all species involved from host separation, some of the specimens still could be a product their entire geographical distribution as well as by obtaining of multiple individual seedlings infecting a plant (Costea and additional critically needed corroborating information. Thus, Tardif 2006), and we considered these to be Bbulked^ individ- the objectives of this study are (1) to analyze new nrITS and uals. Therefore, when available, for a number of samples, we trnL-F sequence data to further investigate if the initial topo- collected seed as well. After scarification in concentrated sul- logical conflict is maintained with the expanded sampling, (2) furic acid for 2–5minandmultiplerinsesindistilledwater,we obtain for the first time in this group information about chro- germinated seed on wet filter paper in Petri dish. DNA was mosome numbers from multiple individuals of all species extracted from individual seedlings as well as a mix of seed- involved to find out if karyotype data support the hybridiza- lings, to increase the overall amounts of DNA (Appendix 1). tion hypothesis, (3) investigate the morphological variation of On the phylogenetic trees, uppercase letters after the DNA the group thorough explicit morphometric analyses, and (4) accession number indicate individuals growing on different compile additional data about ecology, host range, and geo- host plants in natural populations, whereas lowercase letters graphical distribution using both herbarium and newly col- indicate individual seedlings from the same mother plant. A lected specimens for this study. We also describe and illus- total of 97 ingroup samples were used for the molecular anal- trate one new allotetraploid species that belongs to this clade/ yses in this study. Based on our previous, more inclusive phy- section. logenetic analyses of Cuscuta subg. Grammica (Stefanović et al. 2007; Stefanović and Costea 2008), and the whole genus (García et al. 2014 ), Cuscuta compacta Juss. ex Choisy, a Materials and methods species belonging to the sister section Oxycarpae, was selected as the outgroup. Taxon sampling A subset of 24 samples was also used for chromosome counts: 10 of C. denticulata,7ofC. nevadensis,3of As part of our overall efforts to elucidate evolutionary history C. psorothamnensis,and4ofC. veatchii (Appendix 1). and biology of Cuscuta, and in preparation for a future mono- graph of the genus, we have searched for relevant specimens Molecular techniques in over 100 herbaria over the years. Approximately 400 herbarium specimens were identified, annotated, and examined Sequences for the internal transcribed spacer (ITS) region of for basic morphology, as well as host and geographical range nuclear ribosomal DNA (nrDNA) as well as trnL-F intron/ for species in Cuscuta sect. Denticulatae. In addition, we con- spacer region from the plastid genome (ptDNA) were obtain- ducted a series of targeted field trips to the areas of particular ed to infer phylogenetic relationships among species of sect. interest for this section in the springs/summers of 2013–2016. Denticulatae. DNA extractions, polymerase chain reaction Efforts were made to ensure a sampling from localities across (PCR) reagents and conditions, and amplicon purifications the entire known geographical range for each of the species followed the protocols detailed in Stefanović et al. (2007). (Costea et al. 2005). Cleaned products were sequenced at the McGill University From these collections, a total of 90 specimens were select- and Génome Québec Innovation Centre (Canada). By direct ed for the morphometric analyses: 37 for C. denticulata,25for sequencing of nrITS amplicons using Sanger sequencing, sig- C. nevadensis,22forC. veatchii, and 6 for the new species nificant amounts of additive polymorphic sites were detected described here, Cuscuta psorothamnensis (Appendix 1). for specimens collected from the Anza-Borrego Desert State Compared to our previous studies (Stefanović et al. 2007; Park, CA, USA, originally identified as C. denticulata,but Stefanović and Costea 2008;Garcíaetal.2014), we substan- henceforth referred to in this study as Cuscuta tially improved our population-level sampling across all spe- psorothamnesis. Purified PCR products were cloned for all cies of Cuscuta sect. Denticulatae, representing the breadth of the species using the pGEM-T Easy Vector II cloning kit their morphological diversity and geographical range (Fig. 1). (Promega) and multiple clones per individual were sequenced. In addition to the eight DNA samples used previously, total Cloning of nrITS amplicons was performed also for one or genomic DNAwas isolated from 41 newly obtained localities, two individuals of all the species even when no polymor- coming from additional herbarium material and field- phisms were detected by direct sequencing. To reduce unin- collected (Appendix 1). Herbarium-derived samples repre- formative repetition in both nrITS and trnL-F matrices, and to sented a single individual per population (samples circled in reduce the computational burden, all the individuals from the Fig. 1). To account for both intra-specific and potential intra- same locality that had identical sequences were grouped into a 386 García M.A. et al. single operational taxonomic unit (OTU). In this fashion, in- with MULTREES on and MaxTrees set to one million. cluding the outgroup, a total of 115 nrITS and 51 trnL-F Support for clades was inferred by nonparametric sequences were analyzed (Table 1). New sequences generated bootstrapping (Felsenstein 1985), using 500 heuristic boot- for this study (106 nrITS and 45 trnL-F)weredepositedin strap replicates, each with 20 random addition cycles, TBR GenBank (accession numbers MH923079–MH923184 for branch swapping, and MULTREES option off (DeBry and ITS and MH920261–MH920306 for trnL-F; see Appendix Olmstead 2000). Nodes receiving bootstrap (BS) values < 1). Sequences from the outgroup species, C. compacta,were 60, 60–75, and > 75% were considered weakly, moderately, newly obtained for this paper and also submitted to GenBank and strongly supported, respectively. with numbers MH920312 (nrITS) and MH920261 (trnL-F). Bayesian phylogenetic inferences were performed using MrBayes v.3.2.6 (Ronquist et al. 2012) run on the CIPRES Sequence alignment and phylogenetic analyses Science Gateway (Miller et al. 2010). The program MrModeltest v.2.3 (Nylander 2004) was used to determine Sequencher 4.2 (Gene Codes Corp., Ann Arbor, MI, USA) the model of sequence evolution for each dataset by the was used to assemble and edit chromatograms of complemen- Hierarchical Likelihood Ratio Tests (hLRTs) and the Akaike tary strands. Sequences were aligned manually using Se-Al Information Criterion (AIC). For the sequence partition of the v.2.0a11 (Rambaut 2002). Gaps were manually coded as sim- nrITS matrix, the Hasegawa-Kishino-Yano model of DNA ple indels (Simmons and Ochoterena 2000) and appended to substitution (Hasegawa et al. 1985) with addition of rate var- the sequence matrices as binary characters. Three gaps were iation among nucleotides following a discrete gamma distri- coded for the nrITS matrix and eight for the trnL-F matrix, bution (HKY + G) was selected as the best fit. For the trnL-F whereas all the gaps were also included in the combined ma- matrix, the General Time Reversible model (Tavaré 1986) trix. Phylogenetic analyses were conducted under parsimony with addition of rate variation among nucleotides following and Bayesian optimality criteria; summary descriptions of a discrete gamma distribution (GTR + G) was the model cho- these analyses, for individual as well as combined datasets, sen; see Table 1 for details. In all cases, the restriction 0/1 are provided in Table 1. state model was selected for the indel partitions. Each Under parsimony criterion, nucleotide characters were Bayesian analysis consisted of two runs, each for 10 mil- treated as unordered and all changes were equally weighted, lion generations starting from a random tree using the de- except for the indel partition which was double weighted. fault priors, and eight Markov chains sampled every 5000 Searches for most parsimonious (MP) trees for all the ma- generations. Of the trees obtained from the two runs, the trices were performed using a two-stage strategy using first 25% were discarded as burn-in. In all analyses, the PAUP* v.4.0a147 (Swofford 2002). First, the analyses in- standard deviation of split frequencies was below 0.01 as volved 10,000 replicates with stepwise random taxon addi- indication of convergence. The 50% majority-rule consen- tion, tree bisection-reconnection (TBR) branch swapping sus trees and the Bayesian posterior probabilities (PP) were saving no more than 10 trees per replicate, and obtained in MrBayes from the 3002 remaining trees. Only MULTREES off. The second round of analyses was per- the nodes receiving ≥ 0.95 PP were considered statistically formed on all trees in memory with the same settings except significantly supported (Rannala and Yang 1996).

Table 1 Summary descriptions for sequences included in Nuclear (ITS) Plastid (trnL-F) Combined phylogenetic analyses and trees derived from individual and Number of OTUs included 115 51 38 combined datasets of Cuscuta Sequence characteristics sect. Denticulatae Aligned length 670 547 1215 Number of indels coded 3 8 11 Variable sites 250 89 188 Parsimony informative sites 93 52 98 MP tree characteristics Length 383 108 231 CI/RI 0.802/0.964 0.935/0.990 0.931/0.988 Bayesian analyses Model of DNA evolution HKY + G GTR + G HKY + G/GTR + G Mean -lnL 3299.63 1294.32 2759.57

CI consistency index, RI retention index Cladogenesis and reticulation in Cuscuta sect. Denticulatae (Convolvulaceae) 387

To conduct a total-evidence analysis, a matrix of equipped with a Pax-it 8 imaging software. For scanning elec- concatenated nrITS, trnL-F, and indel partitions was created tron microscopy (SEM) of pollen, we used hexamethydisilazane for the individuals and populations of the putative parental (HMDS) as an alternative for critical dry point (Costea et al. species (C. denticulata and C. nevadensis) but excluding 2011a, b), and the examination was done at 10 kV using a C. veatchii and the individuals of C. psorothamnensis.This Hitachi SU1510 variable pressure scanning electron micro- matrix was analyzed using both parsimony and Bayesian in- scope. To determine the extent of morphological variation, the ferences following the same procedures than for individual data were visualized with both clustering and ordination datasets (see Table 1). methods using PAST (version 3.15; Hammer et al. 2001). Principal Coordinate Analysis (PCoA or Metric Chromosome techniques Multidimensional Scaling) and Unweighted Pair-Group Average (UPGMA) were both conducted using the Gower’s Flower buds were collected in the field (Appendix 1) and coefficient of similarity. fixed in Carnoy’s solution (absolute ethanol/glacial acetic acid, 3:1, v/v), stored at − 20 °C, stained in darkness with 4% Hosts and geographical range Wittman’s hematoxylin for at least 24 h, and mounted and squashed in 45% acetic acid. Alternatively, stems, seedlings, The geographical distribution, phenology, elevation, and host or flower meristems (grown from seed in the University of ranges are based on observations made in the field and from Toronto Mississauga greenhouse) were prefixed in 8- the following herbaria: ARIZ, ASU, BRIT, CAS, CHSC, hydroxyquinoline 0.002 M for 24 h at 10 °C, fixed in CIIDIR, CIMI, DS, F, GH, IEB, JEPS, K, MEXU, MICH, Carnoy’s solution for 24 h at room temperature, hydrolyzed MO, NY, RSA, SD, TRTE, UC, US, UCR, and WLU. Host in HCl 5 N for 20 min, washed in distilled water, and squashed ranges of the four species (OTUs defined as for morphometric and mounted in 45% acetic acid. The slides were frozen in studies) were initially analyzed with PCoA which indicated liquid nitrogen to remove the coverslip, air-dried and chromo- strongly divergent patterns among the four OTUs (results somes stained with 1% hematoxylin for a few seconds, not shown). The host range was visualized as a bipartite net- washed with distilled water, air-dried, and permanent mounts work in Fig. 5 and summarized in Suppl. Table S1. prepared in Canada balsam.

Morphometric analyses Results

Four OTUs corresponding to C. denticulata, C. nevadensis, Phylogenetic analyses C. veatchii,andC. psorothamnensis were included in the morphometric analyses to test their morphological distinctive- Summary descriptions for sequences obtained from nrITS and ness. Samples collected in the field (also used for molecular trnL-F regions are presented in Table 1. Overall phylogenetic analyses; Appendix 1) were stored in 50% ethanol prior to relationships in sect. Denticulatae are summarized in Fig. 2 their use for morphological measurements. Flowers removed (compare with Suppl. Figs. S1 and S2 for details). from herbarium specimens were steeped in gradually warmed Direct sequencing of nrITS amplicons from most of the 50% ethanol, which was then allowed to boil for a few sec- individuals of the three traditionally accepted species showed onds to rehydrate tissues. A previous morphological study of no intra-individual sequence variation or clear polymorphisms sect. Denticulatae (Costea et al. 2005) and morphometric were limited to only a few positions. This reduced intra- studies of other groups within subg. Grammica (e.g., sect. individual nrITS variation was confirmed by cloning in which Californicae, Costea et al. 2009;sect.Cleistogrammica, the number of variable sites ranged from 0 to 8 in Costea et al. 2015b) provided a preliminary list of useful char- C. denticulata, 0 to 13 in C. nevadensis and 0 to 5 in acters. These characters were further refined using some re- C. veatchii. However, the individuals of what we originally cent studies on character evolution of gynoecium and perianth identified as C. denticulata but we now treat and describe here (Wright et al. 2011, 2012) and infrastaminal scales (Riviere as a new species, C. psorothamnensis, showed a comparative- et al. 2013). Pollen length and width were also measured in all ly very high number of variable sites. Cloning of these the specimens to determine if a relationship exists between amplicons confirmed a high intra-individual variation of pollen size and chromosome number. In total, 32 characters, nrITS sequences ranging from 2 to 52 nucleotide positions. 29 continuous and three binary, were used in the morphomet- Phylogenetic analyses of the nrITS sequences resolved two ric analysis (Appendix 2). strongly supported clades, labeled here as D and N (Suppl. For basic morphology, flowers were dissected under a Fig. S1).Oneofthem,theDclade(BS=79;PP=0.94), Nikon SMZ1500 stereomicroscope and imaged with contained all the sequences of C. denticulata, including those PaxCam Arc digital camera (MIS Inc. 2017, Villa Park, IL) found to be autotetraploid (accession number 1398; multiple 388 García M.A. et al.

autotetraploid 1398), C. veatchii, and C. psorothamnensis. The sequences of these two species together with those of individuals of C. denticulata collected predominantly from its southern parts of distribution (open circles; Fig. 1)are resolved in a moderately to strongly supported internal clade (BS = 85; PP = 1.00). The rest of the sequences, including the four accessions that formed a strongly supported clade on the ITS tree, are not resolved as a lineage here. The N clade includes only sequences of C. nevadensis.Withinthis clade, the eight accessions that were resolved as monophyletic on the ITS tree are not resolved as such on the trnL-F trees. The rest of the accessions, however, are on the plastid trees strongly supported as monophyletic (BS = 100; PP = 1.00). This clade includes individuals collected in the eastern parts of distribution of this species (Death Valley, along the border between California and Nevada; eastern parts of Inyo Co., CA, and western parts of Nye Co., NV; solid squares; Fig. 1). Fig. 2 Schematic overview of the phylogenetic relationships in Cuscuta The combined analyses of the matrix, excluding C. veatchii sect. Denticulatae derived from plastid (trnL-F) and nuclear (nrITS) se- and C. psorothamnesis, recovered trees with BS = 100 and quence data. For simplicity, only the moderately to strongly supported PP = 1.00 support for the two main clades, D and N (Suppl. backbone nodes are shown as resolved. Unresolved groups are represented with a box symbol. For full details, compare with Suppl. Figs. S1 and Fig. S3). Within the N clade, the lineages recovered previous- S2. Symbols are the same as described in Fig. 1 ly with either ITS or trnL-F individual matrices are here recovered also but with high support. The sub-clade with eight accessions recovered with ITS, in the combined analyses re- clones from multiple individuals collected at this locality), as ceived BS = 86 and PP = 0.99, and corresponds with popula- well as a subset of clones obtained from C. psorothamnensis. tions subsequently called as N1 (represented with open The rest of the clones from the same specimens of this species squares; Fig. 1). The rest of the accessions of C. nevadensis were resolved in the N clade (BS = 84; PP = 1.00) together were resolved in a second sub-clade (BS = 86; PP = 0.99), with C. nevadensis and C. veatchii. All the individuals of called N2 (represented with solid squares; Fig. 1). Within C. psorothamnensis had at least one clone resolved in each the D clade, the accessions of C. denticulata recovered as of the two major clades. The clone sequences of these individ- monophyletic on the trnL-F trees are in the combined analyses uals had longer branches, especially those found in the clade of also resolved in a sub-clade with high support (BS = 96; PP = C. nevadensis (Suppl. Fig. S1). While these two major clades 1.00). These accessions are called D2 (represented with open (D and N) were strongly supported, internal resolution within circles; Fig. 1). The remaining accessions of C. denticulata are them was very low. For the group of C. denticulata,twosub- called D1 (represented with solid circles; Fig. 1). clades were recovered only under Bayesian analyses but lacking any significant support, whereas a polytomy was recov- Karyotypes ered under parsimony. Only one sub-clade, including four accessions of C. denticulata [165, 485, 1144, and 1473], was Chromosome counts are listed in Appendix 1.Individualsof recovered by both analyses with strong support (BS = 94; C. denticulata are generally diploids with 2n =30 PP = 1.00). As for the group of C. nevadensis, two clone se- monocentric chromosomes and symmetrical karyotype quences from C. psorothamnensis were successively sister to a (Fig. 3a). Interphase nuclei are reticulate, with a few small polytomy with the rest of the sequences in which a clade with chromocenters. One individual of this species (DNA acces- eight accessions of C. nevadensis [476, 585, 1145, 1399, sion number 1398) from the Walker Pass, Kern Co., CA, is a 1409, 1427, 1429, 1480] was recovered with strong support tetraploid with 2n = 60 chromosomes (Fig. 3b). Chromosome in both analyses (BS = 94; PP = 1.00). morphology and size of this individual were similar to those of Parsimony and Bayesian analyses of the trnL-F matrix re- C. denticulata diploids. This individual showed morphologi- sulted in more resolved and internally better supported trees cal features similar to other specimens of C. denticulata but (Suppl. Fig. S2). Two major clades, D and N, were recovered with larger flower parts (see section on morphological charac- again, but this time with very high support (BS = 100, PP = ters). For some specimens of C. nevadensis, chromosomes 1.00 and BS = 99, PP = 1.00, respectively). The D clade con- were not easy to observe, and counts were between 2n =28 tains all the accessions of C. denticulata (including and 2n = 30. Nevertheless, when good metaphases were Cladogenesis and reticulation in Cuscuta sect. Denticulatae (Convolvulaceae) 389

Fig. 3 Karyology of Cuscuta sect. Denticulatae.Mitotic metaphases and interphase nuclei of a diploid C. denticulata (2n = 30), b tetraploid C. denticulata (2n =60),c C. nevadensis (2n = 30), d C. veatchii (2n =60),ande C. psorothamnensis (2n =60).f Diakinesis of C. veatchii showing n = 30 bivalents. Note the scale bar is the same across and corresponds to 5 μm

found, we always counted 30 chromosomes. Based on this, because of their larger flowers. Field-collected samples we consider C. nevadensis to be a diploid with 2n =30. (Stefanović SS-13-33 A and B) have been determined to be Chromosomes of C. nevadensis are monocentric, the karyo- autopolyploids (see BChromosome techniques^ section), but type is symmetrical, and the interphase nuclei are areticulate, no chromosome data is available for herbarium collection exhibiting large and well-defined chromocenters (Fig. 3c). Henrickson 17713. Also, chromosomes in this species are larger than those of The dendrogram obtained from the UPGMA cluster anal- C. denticulata (Fig. 3c; note that all images in this panel are ysis revealed also three distinct clusters which had a similar shown at the same magnification). composition to the major groups obtained through PCoA anal- All the individuals of C. veatchii sampled are tetraploids ysis: one cluster that contained all C. denticulata (including with 2n = 60 chromosomes (Fig. 3d). Whereas in the diploids autotetraploids), one that comprised both C. veatchii and all the chromosomes were of similar size, C. veatchii has C. psorothamnensis,andoneforC. nevadensis (Suppl. Fig. metaphase chromosomes with a mix of different sizes. 4). The six specimens attributed to C. psorothamnesis formed Interphase nuclei in this species have a mixture of small and two sub-clusters within C. veatchii (Suppl. Fig. 4). Within well-delineated chromocenters. The three individuals of each species, in general, samples from the same geographical C. psorothamnensis included in our karyological studies were areas did not cluster together. The cophenetic correlation co- also tetraploids (2n =60;Appendix 1) with similar karyotype efficient was 0.8438. and interphase nuclei to C. veatchii (Fig. 3e). In both tetraploid Analyzed independently, pollen data did not permit the species, meiosis was regular with n = 30 bivalents (Fig. 3f). separation of any of the four OTUs when using either ordination or clustering methods (results not shown), but some Morphometric analyses trends were observed. In general, C. veatchii has the largest pollen grains, while C. nevadensis the smallest. However, With or without the pollen size included in the analyses, PCoA there is a significant overlap of the pollen size especially be- produced three distinct groups: one that corresponded to tween C. denticulata and C. nevadensis. The known autopoly- C. denticulata, one for C. nevadensis, and one for overlapping ploid of C. denticulata (Stefanović SS-13-33;KernCo.,CA) clusters of C. veatchii and C. psorothamnensis (Fig. 4). The as well as suspected autopolyploid (Henrickson 17713;Inyo first coordinate axis (62.621% of the variance) separated Co., CA) have large pollen grains, similar in size to those of C. nevadensis from C. denticulata and C. veatchii/C. C. veatchii (results not shown). C. psorothamnensis grains psorothamnesis mix. The second coordinate axis (9.297% of were either similar to C. veatchii or they were mixed with the variance) clearly separated C. denticulata from C. veatchii those of C. denticulata and C. nevadensis. The shape of pollen and C. psorothamnesis. Two specimens of C. denticulata— grains varied within each OTU from sub-sphaeroidal to pro- Stefanović SS-13-33 A and B (Kern Co., CA) and Henrickson late. The ploidy level does not affect the number of colpi as all 17713 (Inyo Co., CA)—diverged from their species group species/accessions have 3(-4)-zonocolpate pollen. 390 García M.A. et al.

Fig. 4 Principal Coordinate Analysis (PCoA) using all morphological psorothamnesis mix. Circles = C. denticulata;squares=C. nevadensis; characters (Appendix 2). The first coordinate axis (62.621% of the vari- triangles = C. veatchii;stars=C. psorothamnensis. Head arrows indicate ance) separated C. nevadensis from C. denticulata and C. veatchii togeth- the proximity of SS-13-33 (2n =60)and Henrickson 17713 (presumed er with C. psorothamnesis. The second coordinate axis (9.297% of the autotetraploid) variance) clearly separated C. denticulata from C. veatchii/C.

Morphology of the tectum is also similar among OTUs: im- host for C. denticulata, C. nevadensis,andC. psorothamnensis, perforate or with a few isolated punctae, sexine scabrate with but while the latter grows only on P. schottii, the two former may isolated granule. both parasitize Psorothamnus arborescens (A.Grey) Barneby and Psorothamnus spinosus (A.Grey) Barneby and have never been Host range observed on P. schottii (Fig. 5 and Suppl. Table S1).

Host ranges of Denticulatae species are largely distinct from one another (Fig. 5 and Suppl. Table S1). Cuscuta denticulata Discussion parasitizes the largest number of species, 35 from 12 families while C. nevadensis grows on 15 species from 6 families. Natural hybridization in Cuscuta sect. Denticulatae Favorite hosts for C. denticulata appear to be Larrea tridentata J.M.Coult. (Zygophyllaceae) (28.97% frequency) Figure 6 provides a schematic overview of phylogenetic rela- and various species of Chrysothamnus (Asteraceae (totaling tionships among species of Cuscuta sect. Denticulatae, in- 27.08% frequency). For C. nevadensis, Atriplex confertifolia cluding reconstruction of cladogenesis and reticulation in this S.Watson (Chenopodiaceae) (37.77%), Ambrosia dumosa group. The initial phylogenetic evidence on the hybrid origin (A.Grey) W.W.Payne (Asteraceae) (17.67%), and of C. veatchii found by Stefanović and Costea (2008)isnow Psorothamnus fremontii (A.Grey) Barneby (Fabaceae) reconfirmed and is further supported when our extensive sam- (11.11%). The host ranges of C. denticulata and pling is included in the morphological and molecular analyses. C. nevadensis overlap with low frequency on six hosts from Concerns about potential topological artifacts, a simple NNI five families and their most commonly shared host is topological distortion, that might be derived from a limited A. dumosa (3.73% for C. denticulata and 17.67% for sampling of only three ingroup species and one outgroup are C. nevadensis). However, neither in the field nor among her- here alleviated. As previously proposed, our data support barium specimens have we ever encountered any two of these C. denticulata as the maternal and C. nevadensis as the pater- species growing together on the same host plant. Most notably, nal taxa. In spite of the intensive sampling and cloning, in- the two hybrid allopolyploid species, C. veatchii and cluding multiple individuals of C. veatchii,noneorveryfew C. psorothamnensis, are highly host-specific, parasitizing exclu- nrITS polymorphisms were detected, suggesting a complete sively on Pachycormus discolor and Psorothamnus schottii homogenization by concerted evolution to the rDNA type of (Thor.) Barneby, respectively. Genus Psorothamnus is a common C. nevadensis. The populations of C. psorothamnensis, Cladogenesis and reticulation in Cuscuta sect. Denticulatae (Convolvulaceae) 391

Fig. 5 Host ranges of Cuscuta sect. Denticulatae species visualized as a bipartite network. Four Cuscuta species nodes on the left are connected with the corresponding nodes of their hosts on the right. Thickness of the lines representing the edges of the network indicates the frequency of the parasite-host as- sociation. Frequencies higher than 10% are indicated in the graph while the rest of the host frequencies are available in Suppl. Table S1. Hosts indicated with gray dots on white are shared between C. denticulata and C. nevadensis. Note the 100% host specificity of C. veatchii and C. psorothamnensis

described here as a new species, share morphological char- by its geographical distribution, parapatric in a narrow zone acteristics of C. veatchii (Fig. 4) but are well differentiated along the southern limit of the distribution of C. denticulata from it both geographically and by host specificity (Figs. 1 (Fig. 1). It also grows exclusively on P. schottii. While this and 5). The direction of hybridization is the same as in genus is known as occasional host for both diploid parentals, C. veatchii, but both maternal and paternal nrDNA types this particular species is not parasitized by any of them are present in the genome of C. psorothamnensis,lacking (Fig. 5). concerted evolution. Based on the retention of both nrDNA An older origin of C. veatchii is supported by a combina- types, it appears that this species is the result of a more tion of traits: the complete homogenization of the nrDNA recent and independent hybridization event because time arrays through concerted evolution, its clearly disjunct distri- has not yet been sufficient for the homogenization of the bution in Baja California (Mexico) where neither of the dip- ribosomal arrays. The notion that C. psorothamnensis is loid parentals currently occurs, and by host specificity restrict- probably a more recent allotetraploid is further supported ed to Pachycormus, a monotypic genus also endemic to Baja 392 García M.A. et al.

established polyploids in populations of their diploid pro- D genitors is not necessarily guaranteed. In the absence of uniparental reproduction (e.g., by extensive selfing or apo- DD mixis), rare polyploids may suffer from frequency- dependent disadvantages of being surrounded by closely related diploid individuals (Bminority cytotype exclusion^; Levin 1975; Husband 2000; Coyne and Orr 2004). The ability to colonize environments where parental cytotypes PS are absent can reduce competition and enhance survivor- nuc ship of neopolyploids. In the case of parasitic plants, these nuc novel environments may be new geographic areas but also new ecological niches, manifested primarily as new hosts in these plants. nuc V Cytological evidence of the hybrid origin of C. veatchii and C. psorothamnensis is provided by their tetraploid chromosome number, the presence of a mix of 30 small (supposedly derived from C. denticulata) and 30 bigger (from C. nevadensis) chromosomes, and interphase nuclei with chromocenters smaller than in C. nevadensis and the rest of the chromatin denser than in C. denticulata.Further time N research on the karyotype characterization of diploids and Fig. 6 A summary model of relationships within Cuscuta sect. tetraploids is necessary to reveal patterns that are present in Denticulatae synthesized from all data presented in this study the diploid parentals and maintained in the allotetraploids. (molecular, cytological, and morphological), showing the reconstruction Preliminary results using fluorescent staining (DAPI/CMA) of cladogenesis (thick lines) and reticulation (thin lines). Hypothesized and fluorescent and genome in situ hybridization (FISH, relative time frame is indicated. Abbreviations: D, C. denticulata (DD, autopolyploid); N, C. nevadensis; nuc, nuclear ribosomal arrays GISH) confirm the presence of two sets of chromosomes (biparentally inherited; lost copy shown as a dotted line); PS, in C. veatchii, with banding patterns and morphological fea- C. psorothamnensis; pt., plastid genes (maternally inherited); V, tures of the two diploid species (Ibiapino et al., in C. veatchii preparation).

C. denticulata autotetraploids California, belonging to a family (Anacardiaceae) not parasitized by the diploids. A hypothetical origin of C. veatchii might The tetraploids from the Walker Pass (Kern Co., CA) that have occurred in Baja California Peninsula if the diploids had were identified morphologically as C. denticulata have a D1 a wider distribution range in the past and the hybridization and ribotype, not N2/D2 as in C. veatchii and C. psorothamnensis whole genome duplication (WGD) had taken place some- (Suppl. Fig. 1 and Fig. 1,solidcircles).Also,trnL-F se- where along the current area of distribution of C. veatchii or quences belong to D1 haplotype, not D2 as in the allotetra- Pachycormus. A host shift associated with allopolyploidy ploids. In the case of a hybrid origin of these tetraploid plants, would have enabled C. veatchii to successfully establish in the progenitors involved would have been from different pop- the area followed by the local extinction of the diploids. ulations compared to those involved in the origin of Some of the hosts frequented by the diploids, like Larrea C. veatchii and C. psorothamnensis. Even though there is no tridentata or Ambrosia dumosa, are common in Baja topological incongruence between nrITS and plastid se- California (Wiggins 1980) suggesting that other biological quences for these individuals, a hybrid origin cannot be or ecological factors prevent the presence of the diploids in dismissed because concerted evolution might have occurred, the distribution area of C. veatchii. removing the C. nevadensis arrays and keeping only those of An alternative scenario is the origin of C. veatchii in the C. denticulata. However, we argue that these plants are auto- current area of distribution of one or both diploid progen- tetraploids and not allopolyploids, for a number of reasons: itors and the chromosomic and genomic modifications dur- morphological, karyological, and phylogenetic (Fig. 6). ing the allopolyploidization event allowed a host shift and Morphologically, they have very different features compared the migration to the South. This sympatric formation but to C. veatchii and C. psorothamnensis and are instead very allopatric establishment of polyploids is frequent in angio- similar to a typical C. denticulata (Fig. 4), except for bigger sperms (e.g., Stuessy et al. 2004; Grundt et al. 2005; flowers and pollen, consistent with their polyploid condition. Nakagawa 2006;Loetal.2010). The success of newly While the chromosome number is double, the chromosome Cladogenesis and reticulation in Cuscuta sect. Denticulatae (Convolvulaceae) 393 size and karyotype are uniformly similar to those of allopolyploidy has been proposed as the origin of a few C. denticulata (Fig. 3). Lack of a mix of larger and smaller species in the hemiparasitic genus Castilleja (Tank and chromosomes, as observed in hybrid polyploids (C. veatchii Olmstead 2009). However, the exact parental species im- and C. psorothamnensis), is an additional line of evidence for plicated in the formation of the allopolyploids and their autopolyploidy in these individuals of C. denticulata. cytology were not determined. Tetrasomic segregation and multivalent formation are nor- A relatively simple evolutionary model Cuscuta sect. mally expected in meiosis of autopolyploids, but unfortu- Denticulate as described here (Fig. 6) offers the opportunity nately, we could not observe meiosis from limited samples to compare expression profiles between known diploid par- in hand. Up to this point, the chromosome material has ents and to identify post-polyploid changes in gene co- been obtained only from one locality (Walker Pass, Kern expression using two independent lineages of allopolyploids. Co., CA) but we have studied herbarium specimens mor- The presence of an autotetraploid in this system would allow phologically similar from Inyo Co., CA (Henrickson to take into account the dosage effects with respect to at least 17713), suggesting the existence of additional autotetra- one (maternal) progenitor and hence tease apart the effect of ploid populations, which we hope to include in our future allele numbers from that of homeolog gene silencing. studies. Finally, our preliminary phylogenetic results on Searching for differences in transcription between sequencing and cloning of multiple low copy nuclear C. veatchii and C. psorothamnensis will provide important pentatricopeptide repeat (PPR) genes (García et al. in prep- information on whether evolution repeats itself: to what extent aration) also indicate the presence of subgenomes exclu- there are differences in gene expression between recurring sively from C. denticulata in these tetraploids. polyploids and are those changes predictable or stochastic? The autotetraploid cytotype occurs sympatric with the A promising line of research in this model would also be to diploids but might show an incipient niche separation by compare specifically the haustorial gene expression between growing exclusively on Ericameria (Compositae). As for the more generalist diploids and the very host-specific allote- allopolyploids, the host shift may provide an advantage for traploids. The allopolyploids have the genetic material of dip- newly formed polyploids by reducing competition with loid species that allow them to grow on a diverse range on host adjacent diploid populations through the specialization to species. However, hybridization and/or polyploidy have re- one host which is not at all (or is much less) frequented by sulted in two host-specific lineages that apparently have lost the diploids. Our preliminary field observations suggest the ability to grow on hosts frequented by the progenitors. also a different phenology compared to the diploids, with This kind of comparative study has the potential to provide the tetraploids having a later flowering time. Further re- insights into key genes that determine host preferences in par- search is necessary to document the extent of autopolyploi- asitic plants and improve our understanding of the genetic dy in C. denticulata, including the study of more individ- basis of host switch in these plants. uals from this and other morphologically similar populations together with their host range. Taxonomic treatment Future directions Even though C. psorothamnensis has originated from the Most studies about hybridization and polyploidy, including same parents as C. veatchii, there is a clear differentiation in genetic and genomic aspects of these important evolutionary ecology, host specificity, and distribution between these two phenomena, are limited to a few crop and genetic model sys- species as well as with their diploid parents. Host-switching tems (Soltis et al. 2016). Few studies, however, have exam- has been considered an important driver of speciation in par- ined the ecological context inherent to natural systems to un- asitic organisms, especially in obligate parasitic plants derstand the effects of hybridization and WGD to ecology and (Schneider et al. 2016). The description of C. psorothamnensis physiology of affected plants. More studies on non-model and will allow the easy comparison of genomic, chromosomal, natural systems, in which one could begin to consider ecolog- transposon activation, ecological, and other data in future ical effects as well, are necessary to understand the role of studies. polyploidy in species differentiation and evolution. One such Cuscuta psorothamnensis Stefanović,M.A.García& promising model is Mimulus (Phyrmaceae; Vallejo-Marín Costea, sp. nov. et al. 2015) but currently there are very few other well- Type: USA California; San Diego Co., Anza-Borrego documented natural systems with recurrent allopolyploidy. Desert State Park, Hwy S2, mile 51, Stefanović SS-13-07,21 Cuscuta sect. Denticulatae is, to our knowledge, the first April 2013 (holotype: TRTE, isotypes: WLU, NY) (Fig. 7) case of recurrent allopolyploidy documented for any group Diagnosis: Similar morphologically to Cuscuta veatchii of parasitic plants (Figs. 2 and 6). In the root parasitic but flowers usually possess longer pedicels; calyces are family Orobanchaceae, polyploidy is common and glossy, straw-yellow, and the host is Psorothamnus schottii. 394 García M.A. et al.

Fig. 7 Morphology of Cuscuta psorothamnensis. a Flower, lateral view (note the axillary floral bud indicating the development of a second flower). b Flowers, top view. c Calyces after the removal of corolla. d, e Dissected calyx imaged with both diffuse (d) and transmitted light (e). f Corollas after the removal of calyces. g Dissected corolla. h, i Infrastaminal scale variation, attached to corolla tube (h)and detached from it (i). j Gynoecium. k Indehiscent capsule. l Embryo still surrounded by endosperm; note the globose structure characteristic to sect. Denticulatae. Scale bars = 1 mm except i which is 0.5 mm

Stems: filiform to medium, yellow. Inflorescences:twoto isolated punctae, sexine scabrate with isolated granules; five-flowered cymose clusters or flowers solitary; bracts 1–2at infrastaminal scales 0.8–1.4 mm long, equaling corolla tube, thebaseofpedicelsand0–1 at the base of flowers, 0.8–1.6 × bridged at 0.4–0.7 mm, ovate to oblong, truncate, with 10– 0.6–1 mm, membranous, ovate-lanceolate, margins entire to 16 fimbriae, 0.4–0.6 mm long; styles uniformly filiform, irregular, apex acute; pedicels 0.5–1.8 mm. Flowers:(4–)5- 0.3–0.7 mm long, 1/3–1/2 as long as the ovary; stigmas merous, 2.6–3 mm long, membranous; papillae absent but cells capitate, globose. Capsules: indehiscent, globose-ovoid to of corolla lobe tips dome-like; laticifers evident in the calyx ovoid, 1.1–2×1–1.6 mm, not thickened or risen around and corolla, isolated or in rows of two to three; calyx 1.6– the inconspicuous interstylar aperture, not translucent, 1.9 mm long, straw-yellow, reticulate and glossy, divided ca. surrounded by the corolla. Seeds: one per capsule, globose 2/3 the length, tube 0.6–0.8 mm long, campanulate, as long as to globose-ovoid, 1–1.4 × 0.8–1.4 mm, seed coat cells alve- the corolla tube or somewhat longer, lobes 0.9–1.2 mm long, olate/papillate, hilum region ca. 0.2 mm in diameter, termi- basally overlapping, ovate-triangular, not carinate, margins ± nal, sunken, scar 0.08–0.18 mm, embryo globose-enlarged irregular or with a few teeth, apex acute; corolla 2.4–2.7 mm toward the radicular end. long, white when fresh, cream-yellow when dried, tube 1– Etymology: The specific name stems from the name of the 1.2 mm long, campanulate, lobes 1–1.6 mm long initially erect genus, Psorothamnus, to which the new species is host- later spreading to reflexed, somewhat longer than the tube, specific. ovate, margins entire to irregular, apex acute to subacute, Distribution and ecology: Flowering March–April; eleva- straight; stamens included, shorter than corolla lobes, anthers tion 100–500 m; vegetation of Larrea-Ambrosia co-dominant subround to broadly elliptic, 0.3–0.5 × 0.25–0.4 mm, filaments with indigo bush (P. schottii) on bajadas, fans, and lower 0.2–0.4 mm long; pollen 3(-4)-zonocolpate, subprolate to pro- slopes in Anza-Borrego Desert State Park; host: P. schottii. late, 19–23 × 13–19 μm, tectum imperforate or with a few For specimens examined, see Appendix 1. Cladogenesis and reticulation in Cuscuta sect. Denticulatae (Convolvulaceae) 395

Key to the species of Cuscuta sect. Denticulatae Hwy 178, 3 mi E of Shoshone, 23 June 2013, Stefanović SS-13-28 (seeds only); [#1478, MH920279 1. Cymes glomerulate; floral buds rounded; calyx lobes ob- (a), MH920280 (mix) / MH923110 (a,b,c,d)], Stefanović ovate-orbicular, apically rounded to obtuse, margins den- SS 13–28A, B*(TRTE,WLU),2n =30; Panamint Valley ticulate……………………………………C. denticulata Rd. 0.5 mi W of Ballarat (Ghost town), 24 June 2013, 1. Cymes umbellate; floral buds acute; calyx lobes broadly Stefanović SS 13–31 (seeds only) [#1479, MH920281 ovate to triangular lanceolate, apically acute to acumi- (a,b,c,d,mix) / MH923111 (a,b,c,d)], 2n =30;Death nate, margins denticulate to ± entire ……...…………2 Valley NP, Hwy 267, 16 mi E of Scott’s Castle, 28 2.Cymesof1–2(−5) flowers; flowers (2.8–)3–4(−4.5) mm June 2013, Stefanović SS 13–46 (TRTE) [#1481, long; calyx lobes triangular lanceolate, margins ± entire; MH920282 (a,b,c,d,mix) / MH923112 (b,c,d)], 2n =30; corollatube1.3–2.2 mm long; anthers 0.5–0.8 mm long, Surprise Canyon, 27 March 1972, Wiggins 21769 (DS); 1 filaments shorter than anthers………...... C. nevadensis mi E of Ballarat, 12 May 1947, Munz 11731 (SD, UCR); 2. Cymes of (1–)2–5 flowers; flowers 2–3.2 mm long; Steep canyon leading into Saline Valley, 16 September calyx lobes oblong to broadly triangular-ovate, mar- 1960, Thomas 8904 (SD); Horton Creek, 15 September gins ± irregularly denticulate; corolla tube 0.8– 1962, Wheeler 8206 (UCR); 18 mi S of Olancha, 20 1.6 mm long; anthers 0.3–0.5 mm long, filaments October 1978, Henrickson 17713 (NY); 6 mi NE of equaling anthers...... 3 Little Lake, 12 June 1979, Henrickson 18287 (SD); 3. Pedicels 0.5–1.2 mm; calyx dullish, brownish-yellow; co- Pleasant Canyon, 8 June 1995, White & Wilcox 3251 rolla tube 0.8–1.6 mm long; host Pachycormus discolor (UCR); Kern Co.: Canebrake Valley, 7 mi E of Onyx, 9 (Anacardiaceae) ……...... ….…………….……C. veatchii November 1970, Howell 47558 (QFA) [#1474, 3. Pedicels 0.5–1.8 mm; calyx glossy, straw-yellow; MH920277 / MH923108]; Hwy. 178, 10 and 15 mi W corolla tube 1–1.2 mm long; host: Psorothamnus of Hwy. 14, Walker Pass, 24 June 2013, Stefanović SS-13- schottii (Fabaceae)…..…………C. psorothamnensis 33* (TRTE, WLU) [#1398, MH920271 (B,C) / MH923095 (A), MH923096 (B clone 1), MH923097 (B Acknowledgements The authors thank the curators/directors of the many clone 2), MH923098 (B clone 3), MH923099 (B clone 4), herbaria that made available their specimens for study. We are grateful to MH923100 (B clone 5) MH923101 (B clone 6), two anonymous reviewers who kindly provided comments that improved n the quality of the article. This research was supported by NSERC of MH923102 (C)], 2 =60; Dry creek below the Goat Canada Discovery grants to M. Costea (327013) and S. Stefanović Ranch, 26 November 1969, Twisselmann 16344*(CAS) (326439). [#1473, MH920276 / MH923107]; Mono Co.: Hwy 120, 11 mi W of Benton, 26 Jun 2013, Stefanović SS-13-40 (TRTE) [#1401, MH920273 (B) / MH923104 (A), Appendix 1 MH923105 (B)]; Riverside Co.: Midland Rd. 7 mi N of Blythe, 22 April 2013, Stefanović SS-13-10* (TRTE, Herbarium vouchers used for morphometric, molecular and/or WLU) [#1362, MH920263 (A,B) / MH923080 (A), cytological analyses of Cuscuta sect. Denticulatae.Country, MH923081 (B)], 2n =30; Morongo Wash, 11 October county, locality, date, collector(s), and herbarium acronym 1932, Wolf 4282 (SD); NW end of McCoy Mts., 1 are provided. For material used in molecular analyses, DNA May 1991, Glenn 91–50 (UCR). San Bernardino Co.: extraction (as shown on phylogenetic trees) and GenBank ac- Hwy. 95, 3 mi N of Vidal junction, 22 April 2013, cession numbers are given in square parentheses for trnL-F Stefanović SS-13-11* (TRTE, WLU) [#1363, and ITS separated by a forward slash (/). When multiple se- MH920264 (A,B) / MH923082 (A), MH923083 (B)]; quences were obtained, uppercase letters indicate individuals Hwy. 95, 10 mi N of Vidal junction, 22 April 2013, growing on different hosts in the same population, lower case Stefanović SS-13-12* (TRTE, WLU) [#1364, letters indicate individual or mixed seedlings from the same MH920265 / MH923084], 2n =30;Hwy.95,Essex,23 mother plant, and ITS clones are indicated with numbers (see April 2013, Stefanović SS-13-16 (seeds only) [#1368, Suppl. Figs. 1 and 2). A dash (−) indicates missing data. MH920266 / MH923085]; Hwy. 95, 10 mi W of Essex, Specimens that were used both for molecular and morphomet- 20 mi E of Amboy, 23 April 2013, Stefanović SS-13-17 ric analyses are indicated with an Basterisk.^ For those (TRTE) [#1369, MH920267 / MH923086]; Hwy. 95, vouchers analyzed cytologically, the chromosome counts are 15 mi W of Essex, 15 mi E of Amboy, 23 April 2013, also provided in bold. Stefanović SS-13-19* (TRTE, WLU) [#1371, MH920268 (A,B) / MH923087 (A), MH923088 (B)], 2n =30; 1. Cuscuta denticulata.USAArizona.Mohave Co.: 15 km Amboy Rd., N of Hwy 62 (N of Joshua Tree NP), 23 S of Wikieup, 6 May 1993, Baker et al. 10732 (ASU) April 2013, Stefanović SS-13-21* (TRTE, WLU) [#668, EF194411, EF194628]; California. Inyo Co.: [#1373, MH920269 (A,B,C) / MH923089 (A), 396 García M.A. et al.

MH923090 (B clones 1, 5), MH923091 (B clones 2, 6), 2n =30; Panamint Springs, on Hwy. 190, Death Valley MH923092 (B clones 3, 4), MH923093 (C)], 2n =30; N.P., 23 June 2013, Stefanović SS-13-30 (TRTE, WLU) Hwy. 66, 3 mi E of Amboy, 23 April 2013, Stefanović [#1409, MH920290 / MH923131], 2n =30; Hwy. 395, SS-13-22* (TRTE, WLU) [#1374, MH920270 / Coso Junction (and all along the hwy divide), 24 MH923094], 2n =30; Hwy 127, 30 mi N of Baker, 23 June 2013, Stefanović SS-13-34 (TRTE) [#1429, June 2013, Stefanović SS-13-25 (seeds only) [#1477, MH920294 / MH923134], 2n c. 30; Hwy. 190, 3.5 mi MH920278 (a,b,c,d,mix) / MH923109 (a,b,c,d)], 2n = NE of Olancha, 24 June 2013, Stefanović SS-13-35* 30; 15 mi E from Amboy, 20 May 1920, Munz 4181 (TRTE, WLU) [#1399, MH920286 (B,C) / MH923122 (UCR); Whipple Mts., 16 March 1980, Sanders 1161B (B,C)], 2n c. 30; Hwy. 190, 3.5 mi NE of Olancha, 25 (UCR); 24 May 1980, Sanders 1527 (UCR); E Mojave June 2013, Stefanović SS-13-36 (only seeds) [#1480, Desert, 9 June 1999, Pitzer 3998 (UCR); 16 March 2001, MH920295 (a,b,c,d,mix) / MH923135 (a,b,c,d)]; Hwy. Bristol Lake Basin, Sanders 23777 (UCR); SE end of 127, 5 mi N of Death Valley Junction, 29 June 2013, Bristol Mts., 30 May 2004, Sanders 28080 (UCR); 4 mi Stefanović SS-13-48* (TRTE, WLU) [#1407, MH920289 NE of Amboy, 30 May 2004, Sanders 28076 (UCR); / MH923130]; SanBernardinoCo.:Hwy.178,17miS-SW Mojave Desert, Victor Valley, 11 June 2005, Green s.n. of Trona, 24 June 2013, Stefanović SS-13-32 (seeds only) (UCR); 22 April 2011, Sanders 39285 (UCR, WLU); 22 [#1427, MH920292 (a), MH920293 (b,d,e,mix) / April 2011, Sanders 39286 (UCR, WLU). San Diego Co.: MH923133], 2n =30; 8 mi W of Death Valley Junction, Anza-Borrego State Park, just S of Bow Willow 4 May 1958, Raven 12865 (CAS); Howell 32746 (RSA); Campground, 32.8417 N, 116.2075 W, 10 May 2009, Panamint Mts., 9 July 1974, Thorne et al. 44889 (SD); ca. Cain & Clayton 1113 (SD); Borrego Valley, 7 April 16 mi S-SE of Olancha, 19 October 1978, Henrickson 1940, Howe 991 (SD). Tulare Co.: Lamont Meadow, 17659a (NY); Ridge top NW of Indian Pass, 1 Howell 43513 (CAS). Nevada. Esmeralda Co.: June 1983, Peterson 888 (NY); Slate Range, above the Intersection of Hwys. 264 / 6, 25 June 2013, Stefanović Gold Button Mine, 8 May 1981, Sanders 2094 (UCR); SS-13-39 (seeds only) [#1400, MH920272 / MH923103]; White Mts., 18 June 1984, Morefield 2119s* (NY) [#585, Washoe Co.: Pyramid Lake, Hwy. 448, 8 mi W of Nixon, EF194408 / EF194630]; Slate Range, 8 May 1981, Sanders 20 July 2014, Stefanović SS-13-45* (TRTE) [#1405, et al. 2091 (UCR); Kern Co.: 6 mi E of Inyokern, 10 MH920274 / MH923106]; Humboldt Co.: Calico Mts., July 1945, Wheeler 6087 (UCR). 3.5 mi SW of Trona, 1 foothill area W of main Gerlach-Soldier Meadow Rd., 6 July 1983, Thorne et al. 56068 (SD). Nevada. Nye Co.: July 2000, Tiehm 13319* (ASU) [#485, EF194410 / Hwy. 373, intersection with Rd. to Ash Meadows, 29 EF194627]; NNW of Bog Hot Ranch, 12 August 1978, June 2013, Stefanović SS-13-47* (TRTE, WLU) [#1406, Tiehm & Rogers 4724 (NY) [#1472, MH920275, −]. MH920287 (B,C,D), MH920288 (E) / MH923123 (B), Utah. Grand Co.: E side of Ida Gulch, SW of Professor MH923124 (E clone 1), MH923125 (E clone 2), Valley, 21 September 1985, Franklin 2601 (NY) [#1144, MH923126 (E clone 3), MH923127 (E clone 4), MH920262, MH923079]; St. George, 1874, Parry 205 MH923128 (E clone 5), MH923129 (E clone 6)], 2n = (Isotype, NY). Washington. Franklin Co.: Hanford Site, 30; Amargosa, 19 June 1969, Beatley 9055 (RSA); Wahluke Wildlife Area, White Bluffs, 19 July 1994, Beck Steward Valley, 12 April 1970, Beatley 9954 (RSA). &Caplow94051(WTU) [#165, EF194409 / EF194626]. 3. Cuscuta psorothamnensis. USA California. Imperial 2. Cuscuta nevadensis. USA California. Inyo Co.: Co.: Near I-8 in sandy flat at mouth of In-Ko-Pah Westgard Pass, 8.5 mile W of White Mt. road, 18 Gorge, Devil’s Canyon along Myer Creek, 17 April June 1963, Lloyd 2639 (NY, UC) [#1145, MH920283 / 1981, Yatskievych 81–119 (ARIZ) [#555, MH920300 / MH923113]; Death Valley NP, Wildrose campground, 13 MH923168 (clone 1), MH923169 (clones 2,3), April 1968, Pinkava et al. 12181 (ASU) [#476, EF194407 / MH923170 (clone 4), MH923171 (clone 5), MH923172 EF194629]; Hwy. 127, 45 mi N of Baker, 23 June 2013, (clone 6), MH923173 (clone 7)]; San Diego Co.: Anza- Stefanović SS-13-26* (TRTE, WLU) [#1426, MH920291 Borrego Desert State Park, 3.2 air mi SW of Hwy 78/Split (A) / MH923132 (A, C)]; Hwy. 178, 1 mile E of Shoshone, Mt. Rd., 3 March 2010, Hendrickson et al. 4502*(SD) 23 June 2013, Stefanović SS-13-27* (TRTE, WLU) [#1510, MH920297, −]; Lois Neyenesch Folly private [#1396, MH920284 (B) / MH923114 (A), MH923115 property, just NW of Anza-Borrego Desert State Park (B), MH923116 (B clones 1,5), MH923117 (clone 2), boundary, 1 mile W of Hwy S2 at Cranebrake Canyon MH923118 (clone 3), MH923119 (B clone 4)], 2n c. 30; Rd., 32.8972° N, 116.2422° W, 20 March 2005, Nenow Hwy. 190, 6 and 7.5 mi W of Death Valley Jct., 23 and Glacy 162* (SD) [#1512, MH920299 / MH923162 June 2013, Stefanović SS-13-29* (TRTE, WLU) [#1397, (clone 1), MH923163 (clone 2), MH923164 (clone 3), MH920285 (A,B) / MH923120 (A), MH923121 (B)], MH923165 (clone 4), MH923166 (clone 5), MH923167 Cladogenesis and reticulation in Cuscuta sect. Denticulatae (Convolvulaceae) 397

(clone 6)]; Anza-Borrego State Park, June Wash, 32.9775° Wiggins, 14856 (GH, WTU) [#559 = #1146, MH920301 N, 116.2475° W, 5 March 2005, Angel 111 (SD); Anza- / MH923174]; 1 mile S of Las Arrastras, 25 March 1960, Borrego State Park, along Rd. S2, mile 51 and 52, 21 Wiggins & Wiggins, 15933 (WTU) [#1147, MH920302 / April 2013, Stefanović SS-13-07* (TRTE, WLU) [#1361, MH923175]; About 11 miles S of Hw1 along road to MH920296 (A,B,C) / MH923136 (A clone 1), MH923137 Cataviña, 29 May 1987, Thorne 62616 (F); Sierra San (A clone 3), MH923138 (A clone 5), MH923139 (A clone Borja, 25 May 1996, Rebman 3189 (SD) 6), MH923140 (A clone 7), MH923141 (A clone 8), MH923142 (B clone 1), MH923143 (B clone 2), Appendix 2. MH923144 (B clone 3), MH923145 (B clone 4), MH923146 (B clone 5), MH923147 (B clone 6), 1. Pedicel length (mm); 2. Flower length (measured from the MH923148 (B clone 7), MH923149 (B clone 8), base of receptacle to tips of straightened corolla lobes, mm); 3. MH923150 (C clone 1), MH923151 (C clone 2), Calyx lobes length (mm); 4. Calyx lobes width (mm); 5. MH923152 (C clone 3), MH923153 (C clone 4), Calyx lobes, width of membranous margin (mm); 6. Calyx MH923154 (C clone 5), MH923155 (C clone 6), tube length (mm); 7. Number of teeth per calyx lobe; 8. MH923156 (C clone 7)], 2n =60; Anza-Borrego Desert Width of overlapping area between calyx lobes (mm); 9. State Park, 3.85 ESE of Hwy. 78/Old Kane Spring Rd., Length of overlapping area between calyx lobes (mm); 10. 33.1302° N, 116.204° W, 11 March 2009, Sullivan 641* Angle formed by margins of calyx lobes at the tips (SD) [#1511, MH920298 / MH923157 (clone 1), (degrees); 11. Calyx area (mm2); 12. Corolla tube length MH923158 (clone 2), MH923159 (clone 3), MH923160 (mm); 13. Corolla lobes length (mm); 14. Corolla lobes width (clone 4), MH923161 (clone 5)]; Hwy S2, mi 53, 6 mi N (mm); 15. Angle formed by corolla lobe margins at the tip of Ocotillo, 4 April 2016, Stefanović SS-16-12 (TRTE), (degrees); 16. Corolla area (mm2); 17. Stamen filaments 2n =60; Hwy S2, mi 47, Mountain Palms Rd., 4 April length (mm); 18. Anthers length (mm); 19. Anthers width 2016, Stefanović SS-16-14 (TRTE), 2n =60. (mm); 20. Infrastaminal scale length (mm); 21. Infrastaminal 4. Cuscuta veatchii. Mexico. Baja California. Close to scales width (mm); 22. Infrastaminal scales, bridge height Bahía de Los Ángeles, 28° 58′ 47.1″ N, 113° 43′ 13.0″ (mm); 23. Infrastaminal scales, length of longest fimbriae W, 298 m; 30 April 2014, Costea s.n.* A, B (WLU) (mm); 24. Infrastaminal scales, width of fimbriae (mm); 25. [#1463, MH920303 (A,B,C,D,E) / MH923176 Number of fimbriae per infrastaminal scale; 26. Length of (A,B,C,E), MH923177 (D clone 1), MH923178 (D clone shortest style (mm); 28*. Receptacle fleshy gelatinous at the 2), MH923179 (D clone 3), MH923180 (D clone 4), base, present (1), absent (0); 29*. Corolla: detaches easily MH923181 (D clone 5)], 2n=60; N of Punta Prieta, 30 from the receptacle/calyx (1), corolla difficult to separate from April 2014, 28° 57′ 25.4″ N, 114° 09′ 40.2″ W, 240 m, the receptacle/calyx (0); 30*. Corolla lobes margin: serrulate 30 April 2014, Costea s.n.* A, B (WLU) [#1464, or denticulate (1), entire to wavy (0); 31. Pollen grains length MH920304 (A,B,C,D,E) / MH923182 (A,B,C,D,E)], (μm); 32. Pollen grains width (μm). 2n =60; Punta Prieta, 11 October 1941, Harbison 6901 (SD); 1.2 mi NW of Punta Prieta, 1 June 1973, Spellenberg et al. s.n. (NY); Close to Bajia de Los Open Access This article is distributed under the terms of the Creative Ángeles, Arroyo El Palmarito, ca. 4.5 km NW of Commons Attribution 4.0 International License (http:// Cataviña, 6 June 1984, Dice 477 (SD); Los Ángeles Bay, creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give 5May1921,Johnston 3439 (GH); 6 May 1921, Los appropriate credit to the original author(s) and the source, provide a link Ángeles Bay, Johnston 3430 (KEW); 31 July 1966, to the Creative Commons license, and indicate if changes were made. Henrickson 2323 (MICH) (RSA*) [#580, EU288339, EU288354]; Ca. 23 mi NW of Bahia de Los Ángeles, 18 February 1979, Devine 508 (CHSC); Cataviña Sur, ca. References 60 km S of Cataviña, 30 April 2014, Costea s.n.*A,B, C, D (WLU) [#1465, MH920305 (A,B,C,D,E) / Costea, M., & Stefanović,S.(2009).Cuscuta jepsonii (Convolvulaceae): MH923183 (A,B,C,D,E)], 2n =60;CataviñaNorte,ca. An invasive weed or an extinct endemic? American Journal of – 5 km N of Cataviña, 30 April 2014, Costea s.n.* A, B, Botany, 96,1744 1750. Costea, M., & Stefanović, S. (2010). Evolutionary history and taxonomy C, D (WLU) [#1466, MH920306 (A,B,C,D,E) / of the Cuscuta umbellata complex (Convolvulaceae): Evidence of MH923184 (A,B,C,D,E)], 2n =60; Vizcaino Desert, ca. extensive hybridization from discordant nuclear and plastid phylog- 11 mi S of Hwy. 1, along the Rd. to Santa Catarina, 29 enies. Taxon, 59,1783–1800. May 1987, Thorne et al. 62616 (F) [#521, EU288338 / Costea, M., & Tardif, F. J. (2006). Biology of Canadian weeds. Cuscuta campestris Yuncker, C. gronovii Willd. ex Schult., C. umbrosa Beyr. EU288353] (NY) [#760, EU288340 / EU288355]; 5 mi ex Hook., C. epithymum (L.) L. and C. epilinum Weihe. Canadian N of Misión San Borjas, 17 May 1959, Wiggins & Journal of Plant Science, 86,293–316. 398 García M.A. et al.

Costea, M., Nesom, G. L., & Tardif, F. J. (2005). Taxonomic status of Press, M. C., & Phoenix, G. K. (2005). Impacts of parasitic plants on Cuscuta nevadensis and C. veatchii (Convolvulaceae). Brittonia, 57, natural communities. New Phytologist, 166,737–751. 264–272. Rambaut, A. (2002). Se-Al sequence alignment editor, version 2.0a11. Costea, M., Wright, M. A. R., & Stefanović, S. (2009). Untangling the Oxford: University of Oxford. systematics of salt marsh dodders: Cuscuta pacifica, a new segregate Rannala, B., & Yang, Z. (1996). Probability distribution of molecular species from Cuscuta salina (Convolvulaceae). Systematic Botany, evolutionary trees: a new method of phylogenetic inference. 34,787–795. JournalofMolecularEvolution,43,304–311. Costea, M., Ruiz, I. G., & Stefanović,S.(2011a).Systematicsof Riviere, S., Clayson, C., Dockstader, K., Wright, M. A. R., & Costea, M. Bhorned^ dodders: phylogenetic relationships, taxonomy, and two (2013). To attract or to repel? Diversity, evolution and role of the Bmost new species within the Cuscuta chapalana complex peculiar organ^ in the Cuscuta flower (dodder, Convolvulaceae): the (Convolvulaceae). Botany, 89,715–730. infrastaminal scales. Plant Systematics and Evolution, 299,529–552. Costea, M., Spence, I., & Stefanović, S. (2011b). Systematics of Cuscuta Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D. L., Darling, A., Höhna, chinensis species complex (subgenus Grammica, Convolvulaceae): S., Larget, B., Liu, L., Suchard, M. A., & Huelsenbeck, J. P. (2012). evidence for long-distance dispersal and one new species. MrBayes 3.2: efficient Bayesian phylogenetic inference and model Organisms, Diversity & Evolution, 11,373–386. choice across a large model space. Systematic Biology, 61,539–542. Costea, M., García, M. A., & Stefanović, S. (2015a). A phylogenetically Schneider, A. C., Colwell, A. E. L., Schneeweiss, G. M., & Baldwin, B. based infrageneric classification of parasitic plant genus Cuscuta G. (2016). Cryptic host-specific diversity among western hemi- (dodders, Convolvulaceae). Systematic Botany, 40,269–285. sphere broomrapes (Orobanche s.l., Orobanchaceae). Annals of Costea, M., García, M. A., Baute, C., & Stefanović, S. (2015b). Entangled Botany, 118, 1101–1111. evolutionary history of Cuscuta pentagona clade: a story involving Simmons, M. P., & Ochoterena, H. (2000). Gaps as characters in sequence- hybridization and Darwin in the Galapagos. Taxon, 64,1225–1242. based phylogenetic analyses. Systematic Biology, 49, 369–381. Coyne, J. A., & Orr, H. A. (2004). Speciation. Sunderland: Sinauer Associates. Soltis, D. E., Visger, C., Marchant, D. B., & Soltis, P. S. (2016). Dawson,J.H.,Musselman,L.J.,Wolswinkel,P.,&Dörr,I.(1994).Biology Polyploidy: pitfalls and paths to a paradigm. American Journal of and control of Cuscuta. Reviews of Weed Science, 6,265–317. Botany, 103, 1146–1166. DeBry, R. W., & Olmstead, R. G. (2000). A simulation study of reduced Stefanović, S., & Costea, M. (2008). Reticulate evolution in the parasitic tree-search effort in bootstrap resampling analysis. Systematic genus Cuscuta (Convolvulaceae): over and over again. Botany, 86, Biology, 49,171–179. 791–808. Felsenstein, J. (1985). Confidence limits on phylogenies: an approach Stefanović, S., & Olmstead, R. G. (2004). Testing the phylogenetic posi- using the bootstrap. Evolution, 39,783–791. tion of a parasitic plant (Cuscuta, Convolvulaceae, Asteridae): García, M. A., & Martín, M. P. (2007). Phylogeny of Cuscuta subgenus Bayesian inference and the parametric bootstrap on data drawn from Cuscuta (Convolvulaceae) based on nrDNA ITS and chloroplast three genomes. Systematic Biology, 53,384–399. trnL intron sequences. Systematic Botany, 32,899–916. Stefanović, S., & Olmstead, R. G. (2005). Down the slippery slope: García, M. A., Costea, M., Kuzmina, M., & Stefanović, S. (2014). plastid genome evolution in Convolvulaceae. Journal of Phylogeny, character evolution, and biogeography of Cuscuta (dod- Molecular Evolution, 61,292–305. ders; Convolvulaceae) inferred from coding plastid and nuclear se- Stefanović, S., Kuzmina, M., & Costea, M. (2007). Delimitation of major quences. American Journal of Botany, 101,670–690. lineages within Cuscuta subgenus Grammica (dodders; Graffis, A. M., & Kneitel, J. M. (2015). A parasitic plant increases native Convolvulaceae) using plastid and nuclear DNA sequences. and exotic plant species richness in vernal pools. AoB Plants, 7, AmericanJournalofBotany,94,568–589. plv100. https://doi.org/10.1093/aobpla/plv100. Stuessy, T. F., Weiss-Schneeweiss, H., & Keil, D. J. (2004). Diploid and Grundt,H.H.,Obermayer,R.,&Borgen,L.(2005).Ploidallevelsinthe polyploid cytotype distribution in Melampodium cinereum and arctic-alpine polyploid Draba lactea (Brassicaceae) and its low-ploid M. leucanthum (Asteraceae, Heliantheae). American Journal of relatives. Botanical Journal of the Linnean Society, 147,333–347. Botany, 91,889–898. Hammer,Ø.,Harper,D.A.T.,&Ryan,P.D.(2001).PAST— Swofford, D. L. (2002). PAUP*. Phylogenetic analysis using parsimony PAlaeontological STatistics, ver.3.15. Oslo: University of Oslo (*and other methods), version 4.0b2a. Sunderland: Sinauer. https://folk.uio.no/ohammer/past/. Accessed 24 April 2018. Tank, D. C., & Olmstead, R. G. (2009). The evolutionary origin of a Hasegawa, M., Kishino, H., & Yano, T. A. (1985). Dating of the human second radiation of annual Castilleja (Orobanchaceae) species in ape splitting by a molecular clock of mitochondrial DNA. Journal of South America: the role of long distance dispersal and allopolyploi- Molecular Evolution, 22,160–174. dy. American Journal of Botany, 96,1907–1921. Husband, B. C. (2000). Constraints on polyploid evolution: a test of the Tavaré, S. (1986). Some probabilistic and statistical problems in the anal- minority cytotype exclusion principle. Proceedings of the Royal ysis of DNA sequences. Lectures on Mathematics in the Life Society. Biological Sciences Series B. London, 267,217–223. Sciences, 17,57–86. Levin, D. A. (1975). Minority cytotype exclusion in local plant popula- Vallejo-Marín, M., Buggs, R. J. A., Cooley, A. M., & Puzey, J. R. (2015). tions. Taxon, 24(1), 35–43. Speciation by genome duplication: repeated origins and genomic Lo, E. Y. Y., Stefanović, S., & Dickinson, T. A. (2010). Reconstructing composition of the recently formed allopolyploid species Mimulus reticulation history in a phylogenetic framework and the potential of peregrinus. Evolution, 69,1487–1500. allopatric speciation driven by polyploidy in an agamic complex in Wiggins, I. L. (1980). Flora of Baja California. Stanford University Press. Crataegus (Rosaceae). Evolution, 64,3593–3608. Wright, M. A. R., Welsh, M., & Costea, M. (2011). Diversity and evolu- Miller, M. A., Pfeiffer, W., & Schwartz, T. (2010). Creating the CIPRES tion of the gynoecium in Cuscuta (dodders, Convolvulaceae) in Science Gateway for inference of large phylogenetic trees. In relation to their reproductive biology: two styles are better than Proceedings of the Gateway Computing Environments Workshop one. Plant Systematics and Evolution, 296,51–76. (GCE) (pp. 1–8). New Orleans, LA. Wright, M. A. R., Ianni, M. D., & Costea, M. (2012). Diversity and Nakagawa, M. (2006). Ploidy, geographical distribution, and morpholog- evolution of pollen-ovule production in Cuscuta (dodders, ical differentiation of Parasenecio auriculata (Senecioneae; Convolvulaceae) in relation to floral morphology. Plant Asteraceae). Journal of Plant Research, 119,51–61. Systematics and Evolution, 298,369–389. Nylander, J. A. A. (2004). Mr Modeltest v2. Program distributed by the Yuncker, T. G. (1932). The genus Cuscuta. Memoirs of the Torrey author. Uppsala: Evolutionary Biology Centre, Uppsala University. Botanical Club, 18,113–331.